The Nim Global Model
The NIM Global Model
Overview:The Nonhydrostatic Icosahedral Model (NIM) is a multi-scale global model developed by Earth System Research Laboratory. It is nonhydrostatic, finite volume, and formulated on an icosahedral grid with height coordinate in the vertical. The use of a nonhydrostatic model is beneficial for prediction of clouds, convection and precipitation; these can be resolved at length scales below 10 km, and are improved by explicit treatment of vertical accelerations. The design of NIM envisages horizontal grid mesh less than 4 km, and 192 vertical levels to resolve shallow clouds and dynamic layers such as the tropopause.
Design Concepts:An important design aspect is the use of three-dimensional control volumes in the finite volume model, with a particularly simple dynamic formulation that uses the five conservation laws in flux form: conservation of momentum in three directions, conservation of thermodynamic energy, and conservation of mass. The use of three-dimensional control volumes allows good conservation properties, with pressure force calculated over terrain based on Newton’s Third Law without the errors associated with large compensating terms. NIM has fully three-dimensional advection, which improves conservation of dynamic fields and tracers. The icosahedral grid has an inherent advantage of having the most uniform mesh size over the globe, an improvement over the commonly used latitude and longitude based grids currently in wide use. The NIM thus includes the advantages of icosahedral models for polar and tropical weather. Other attributes include the A grid, fourth order Runge- Kutta time differencing, and vertically implicit formulation. NIM is similar to ESRL’s other global model, the FIM, but also differs in important respects. FIM is hydrostatic, and uses a hybrid vertical coordinate system. Both models have similar treatment of horizontal dynamics, and share grid formulations.
Computational Design:A primary design requirement for NIM was to be able to take advantage of the new generation of Massively Parallel Fine Grain computers. These computers, such as NVIDIAs GPU and Intel’s MIC, allow a new level of integrated processing, with resultant advantages in faster speeds, lower cost, and lower power consumption. The indirect grid designed by ESRL for models on irregular grids is used.